Seth Walder

28 Mar 2025

Updated NFL analytics staffer list!* As always, this list is to the best of my understanding based both on what teams list and conversations with analytics folks around the league. (*Re-posting because I had a mistake on previous post)

14h

Eagles were playing press coverage on George Pickens early, but did not on either of his two receptions including the TD. Pickens yards per route run this season (including tonight): vs. Press: 0.8 vs. All Other Cushion Types: 2.5 Per Next Gen Stats

15h

Watch the Playbook altcast, on NFL Network! @mikemonaco.bsky.social, Louis Riddick, Brian Hoyer and Brian Burke on tonight. Plus, @benjaminsolak.bsky.social at halftime!!

19h

If NFL MVP was Most Valuable Person -- not just Player -- it would be Brian Flores based on the season thus far. And he was briefly available as a free agent for $0 against the cap this offseason before re-signing with the Vikings.

Worst Field Goals Made Over Expected this season, per NFL Next Gen Stats Harrison Mevis, -2.3 Trey Smack, -2.1 Harrison Butker, -1.4 Riley Patterson, -1.1 Cairo Santos, -0.7

honestly jets are doing a pretty nice job of threading the needle that is: develop young players but lose games so you can get a QB

Solid go on 4th down. This one will be second guessed because of where they are on the field. But value of possession is very high and conversion chance is high on a 4th-and-1. Remember: punts are turnovers ~100% of the time.

Model supported Packers go decision on 4th-and-4

4th down model was clearly in favor of Dolphins decision to go for it on 4th-and-1 with 2:00 left and a shot to end the game.

just a brutal pick by daniel jones to end the colts chances

pretty funny this happened immediately after this week's #discourse

Aaron Schatz@aaronschatz.com

Both of the main fourth down bots said to kick the FG.

Play failed but model supported Ravens' decision to go for it on 4th-and-2.

2 percentage point error by Todd Bowles kicking from the 4-yard line.

Week 7 predictions: upsets, stat lines, fantasy sleepers and more. Alongside Pam Maldonado, Ben Solak, Matt Bowen and Eric Moody! www.espn.com/nfl/story/_/...

Model preferred punt!

Green light go on 4th-and-2 here Breakeven percentage: 46% Conversion probability: 62%

Seth Walder reposted

Rynehawk

@rynehawk.bsky.social

Replying to @sethwalder.bsky.social

Vrabel deserves credit for trying any type of offense at the end of the half! 9 seconds left, two timeouts at -34. A lottttttt of head coaches are kneeling it out there. Rams essentially gave up at that same spot on the field this week with 46 (!) seconds left, 0 timeouts.

Wildest dot path here: Malik Muhammad (24)

Patriots should be calling timeout. Basically no downside.

heck of a job blocking by Rome Odunze on the long Burden reception

Didn't have this in time to get into our TNF betting story for the week but two bets my models like for tonight: Milton Williams over 0.25 sacks (+337) Patriots to record first sack of game (+160)

So @ben-tmeyer.bsky.social made this very fun cool new thing: the NBA Trade Lab! Let's you build fake trades, vote on deals proposed for your team and see deals liked by fans of both teams. If you want this for NFL let us know so we can bully our boss into letting Ben make it!

espnanalytics.comNBA Trade Lab - ESPN AnalyticsBuild, swipe and explore NBA trades. Test ideas against real salary cap rules, vote as a team's GM and see which trades fans like best.

Here it is! espnanalytics.com/nba-trade-lab

espn.comProposing NFL trade deadline offers: Moves for Saquon Barkley, Khalil Mack, moreSaquon Barkley joining a contender? Brian Thomas Jr. headed out west? Khalil Mack and Donte Jackson boosting a defense? Let's build 11 win-win trades.

always read @billbarnwell.com fake trades www.espn.com/nfl/story/_/...

Updated leaderboards for to-be-named cornerback metrics! Quick thread on how these work:

One outstanding challenge is zone coverage when a corner’s receiver responsibility changes mid-play. The ghost/metric does not handle that as well as I would like, but fixes I’ve tried to implement thus far have not helped. This, though, is mitigated by the fact that…

the metric places a heavier weight on man assignment (so still could be man within a Cover-3, for example) snaps vs. zone assignment snaps. This split was determined by strength of correlation with several other measures of cornerback ability.

Updated leaderboards for to-be-named cornerback metrics! Quick thread on how these work:

The combination of Phase 1 and Phase 2 is additive. However, Phase 2 only occurs on targets. So on an untargeted play it is Phase 1 score + 0. On an untargeted play it is Phase 1 + Phase 2. Phase 2 has much more extreme results, but occurs on fewer plays.

Phase 2 is more explanatory of a player’s result, but it is almost completely unstable. Phase 1, however, is quite stable from year-to-year. This last part is remarkable: very few cornerback metrics have much stability at all.

Updated leaderboards for to-be-named cornerback metrics! Quick thread on how these work:

Phase 2 is simpler: it measures the expected yards allowed shortly before the pass is released (roughly start of windup) and compares it to the expected yards allowed at the time of ball arrival (after complete/incomplete but before YAC). It is on targets only.

Phase 2 also has bonuses and debits for interceptions, touchdowns allowed and defensive pass interference (pre-throw penalties like defensive holding and illegal contact are in Phase 1)

Updated leaderboards for to-be-named cornerback metrics! Quick thread on how these work:

Here’s an example of a play where the model believes the cornerback played worse than the ghost. This is Denzel Ward playing soft coverage, therefore boosting the probability of completion to his receiver.

This play was targeted and completed, but the Phase 1 score would be the same regardless of those things.

Updated leaderboards for to-be-named cornerback metrics! Quick thread on how these work:

Phase 1 is a measurement of a cornerback’s positioning – regardless of target – prior to the quarterback’s windup.

Throughout the route it measures the expected yards allowed, assuming target, for both the cornerback and a “ghost defender” in his stead. The Phase 1 score is based on the difference in expected yards between the two.

Playbook is back again tonight! Today it's @fieldyates.bsky.social, Domonique Foxworth, Brian Hoyer and Brian Burke. On NFL Network!

Seth Walder reposted

Aaron Schatz

@aaronschatz.com

Lowest pass block win rate and they allowed no sacks! That's heavily Stroud.

Seth Walder@sethwalder.bsky.social

C.J. Stroud recorded the highest QBR in Week 6 (thus far) while the Texans recorded the lowest pass block win rate in Week 6 (thus far).